Skip to content. | Skip to navigation

Personal tools


You are here: Home / Data / SNooPy / Frequently Asked Questions

Frequently Asked Questions

Here you may find answers to you question about:calvin

Compiling SNooPy
Running SNooPy
Fitting Data

  • Issues Compiling SNooPy
    • While compiling, I get tons of error messages, don't you debug your code?
      Scroll up to where the errors start and check to see if there are any messages about include (*.h) files not being found. You many need to install the proper *-devel.rpm files in Linux (or see below for Mavericks headaches). Or you may be missing a standard library.
    • I'm compiling on my linux system and it says it can't find libg2c or libstdc, aren't those standard libraries?
      If you're using g77 instead of gfortran, the "standard" libraries may be hidden in non-standard places. You can use the following command to figure out where things are:  g77 -print-search-dirs
    • I'm compiling on OSX and I get an error about not being able to find libcc_dynamic.
      You're probably using g77. Make sure you have gfortran and force your system to use it by setting an environment variable:
    • I'm running OSX [Mountain]Lion and I don't have any compilers!  But I installed Xcode.  WTF?
      Launch Xcode, select Preferences... from the Xcode menu, select the Downloads button, select the "Command Line Tools" from the list and click the Install button. Yeah, I know.  But they fixed this in Mavericks (well, almost...  see next item).
    • I'm running on OSX > 10.8 and I don't have any header files! /usr/include is completely missing. Where's my ibuprofen?
      Once you have Xcode installed, open a terminal and run 'xcode-select --install'.  Duh!
    • I get error messages about fortran compilers and it seems to be trying to use /.../iraf/iraf/unix/hlib/ IRAF? Seriously?
      Scisoft, Ureka and other systems with IRAF set up many environment variables, including F77, that can interfere with compiling and running SNooPy (or any other packages requiring FORTRAN). My advice is to disable scisoft by default (comment out the startup script in your shell configuration) and make an alias or something to activate it on-demand. For Ureka, just unsetenv the F77 variable before you run the script.
  • Issues Running SNooPy
    • It takes FOREVER for SNooPy to start up an interactive shell, but scripts importing snpy have no delay. What gives?
      When SNooPy is run interactively, it checks for updates on my repository. If the internet connection is slow (or non-existent), this can take a few seconds. To disable this, run snpy with the -nc switch.
    • I spent all this time getting SNooPy compiled and running. But now I go back, start it up, and it gives me an ImportError, can't find <blah blah blah>. I think I'm going to cry.
      If you followed my advice and installed SNooPy with a separate python distribution, you need to make sure that it is "active" before you start up the snpy script. Depending on the distribution, this might be done by sourcing a setup script or simply making sure the correct python executable is first in your PATH.
    • On OSX SNooPy fails to launch a window when I plot an object, saying it can't find the X server or X libraries.  What gives?
      What gives is that you're running OSX post 10.8 and X-windows is no longer shipped with the system.  You can install it easily, though, by downloading and installing X-quartz.
    • I really dig Gaussian Processes, but SNooPy doesn't provide them as an option for fitting. You lied to me!
      In order to get the Gaussian Processes, you need to install pymc before you install SNooPy.
    • I thought your PGPLOT version of SNooPy was far superior. What would it take to bring it back?
      A time machine.
  • Issues Fitting Data:
    • I get an error messages while fitting saying that the covariance matrix is singular. That sounds bad, what should I do?
      This usually happens when the model becomes insensitive to a parameter or two or more parameters are degenerate. Try fitting while holding different parameters fixed, then freeing them when you get close to a good solution.
    • You know, those parameter uncertainties are awfully small. Who are you trying to fool?
      The errors reported by SNooPy after you have fit the data are purely statistical errors. So if your light-curve has a high density of points with really low photometric errors, the statistical errors in the fit parameters will be very small, even if the fits aren't that great. There are, of course, systematic errors but which ones to use depend on what you are trying to do. After the fit, call the SN object's systematics() function. That will report systematic errors and you can decide which ones you need to include in your analysis.
    • I've got data in a different photometric system, can you help me out?
      SNooPy will do S-corrections automatically for you, if your photometric system is defined. You can follow these instructions for adding your own filter functions to SNooPy.
    • Who uses least-squares anymore? When are you going to implement Bayesian priors?
      SNooPy now has an MCMC fitter that you can use with all your crazy Bayesian hierarchical models. Read about it here.
  • Miscellaneous:
    • Thanks man, SNooPy really helped me out with my work. What can I do for you?
      Please cite my SNooPy paper in your publication:  Burns et al. (2010).  If you use the new color model, also cite Burns et al. (2014).
    • I want to contribute some awesome code I wrote for SNooPy. How do I do that?
      SNooPy is on github. Clone the repository, make your changes, and submit a pull request. Then I'll try to figure out what I'm supposed to do with that.
    • You know, I've got this great idea to make SNooPy way better, are you open to suggestions?
      Sure, but apparently I become more susceptible to suggestions when I've had a good glass of Single Malt.